Getting started

You should be running your first prompt in under two minutes.

1. Install the desktop app

The client is a signed Electron build for macOS, Windows, and Linux. Grab the latest from the download page.

All builds are code-signed. On first launch, macOS will prompt you to allow network access — that's the app fetching the NLP engine address.

2. Create an API key

You need a key to authenticate against the NLP engine. Create one from your dashboard's Keys page. Keys are scoped per device — you can revoke one without invalidating the others.

iom_8f4c...     ← copy this once; it's shown only at creation time

Paste it into the client's Settings → Account screen.

3. Point it at a MongoDB

In Settings → Databases, add a connection string. The client stores it encrypted in your OS keychain (macOS Keychain, Windows Credential Manager, or libsecret on Linux) — never on disk as plaintext.

A read-only role is strongly recommended. See Connecting MongoDB for the minimal Atlas / self-hosted role definitions.

4. Ask a question

Hit ⌘K (or Ctrl+K), type a question in plain English:

Show me users who signed up last week and have more than 2 orders.

Insight O' Mate will:

1. Inspect the schema of your users and orders collections
2. Send (prompt + schema) to the NLP engine
3. Receive a pipeline like db.users.aggregate([{ $match: ... }, { $lookup: ... }])
4. Run it locally and render the result table

You can click the Query tab at any time to see exactly what ran — or copy it into mongosh if you'd rather review before executing.

What to read next

• If the prompt produced the wrong query, skim the Privacy model — it explains what the engine sees and doesn't, which helps you phrase prompts more precisely.
• If you want to script it into a cron job or a CI check, jump to the API reference.